from langchain_huggingface import HuggingFaceEndpoint,ChatHuggingFace
from langchain_core.messages import HumanMessage,SystemMessage
import os
import pandas as pd
from agent.agent_graph.graph import compiled_graph
from agent.rag.rag import rag_text_chooser
import sys
import os
from agent.agent_graph.StateTasks import Available_Tasks
from agent.tools.PDF import PDF_generator_Node
from agent.tools.email import EMAIL_sender_Node
from agent.agent_graph.Graph_Nodes import get_llm_answer
from agent.llm.prompts import NODES_Prompts
import dotenv

sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..', '..')))
dotenv.load_dotenv("/content/drive/MyDrive/study/Projects/keys.env")
def get_response(prompt,memory,hf_key,state,user_email,user_name):
        # Setting up models
        os.environ["HF_TOKEN"]  = hf_key
        llm_gpt = HuggingFaceEndpoint(
            repo_id="openai/gpt-oss-20b",#"deepseek-ai/DeepSeek-V3.2-Exp",#"openai/gpt-oss-20b",
            task='conversational',
            provider="auto",
                max_new_tokens=2048
        )
        llm_gpt = ChatHuggingFace(llm=llm_gpt)

        print("RAG_PATH ",os.path.join(os.path.dirname(__file__), 'agent' ,'rag', 'rag.xlsx'), os.path.exists(os.path.join(os.path.dirname(__file__), 'agent' ,'rag', 'rag.xlsx')))
        
        rag_model = rag_text_chooser(os.path.join(os.path.dirname(__file__), 'agent' ,'rag', 'rag.xlsx'))

        # update state
        state["question"] = prompt
        state["memory"] = memory
        state["llm"] = llm_gpt
        state["rag_model"] = rag_model

        call = compiled_graph.invoke(state)

        save_send_email(call,user_email,user_name)
        os.environ["HF_TOKEN"]  = "" # to prevent keep it in env for other calls and for security

        return call

def save_send_email(call,user_email,user_name):
    if ("all_ok" in call.keys()):
        if (call['all_ok']== True):
          if (call['question_type'] in [Available_Tasks.LAPTOP_CHOOSE.value ,  
            Available_Tasks.QUESTION.value ,
    Available_Tasks.ROADMAP.value]):
              email_txt = get_llm_answer(model_llm=call['llm'],messages=[HumanMessage(content=("اسم الزميل لتستخدمه هو :  "+ user_name +"/n/n")+ NODES_Prompts.Email_text.value + call['question'] + str(call['memory']) +  call['question_type'] + call['answer'])])
              title = get_llm_answer(model_llm=call['llm'],messages=[HumanMessage(content=NODES_Prompts.Email_title.value + call['question'] + str(call['memory']) +  call['question_type']+ call['answer'])])

              import tempfile
              path_pdf = ''
              with tempfile.NamedTemporaryFile(suffix=".pdf", delete=False) as tmp_file:
                  path_pdf = tmp_file.name
                  # هنا تكتب الكود اللي بيولد الملف
                  print("PDF path:", path_pdf)

              # بعد ما تخلص من الملف ممكن تحذفه
              # import os
              # os.remove(path_pdf)


              #path_pdf ="/content/drive/MyDrive/study/Projects/CodeBuddyAI/tmp.pdf"
              PDF_generator_Node(call['answer'],title,path_pdf)
              #EMAIL_sender_Node(user_email,email_txt,title,path_pdf)
              import os
              os.remove(path_pdf)
              print("Done")